Your teams already use AI. But nobody governs it. Voban gives you visibility on what data flows through which models, and lets you enforce your own rules.
| Department | Model | Hosting | Guard rules | Requests (30d) |
|---|---|---|---|---|
| Legal | Mistral Large | EU | PII masking | |
| Engineering | Claude Sonnet | US · EU region | Injection detect | |
| HR | Llama 3.3 | EU | PII + routing | |
| Marketing | GPT-4o | US · EU region | Logging | |
| Finance | Qwen 2.5 | EU | PII masking |
The problem
Some teams use AI tools without approval. Others have official tools deployed, but no visibility on what data goes through them, no rules enforced. You checked the "AI tools" box, not the "AI governance" box.
Voban
Voban gives your teams access to the best AI models — any model, any provider — inside a framework you define. You choose which models to activate, what security rules apply, how data is handled. The platform is operated in Europe.
Guard
Every request goes through the Guard — a configurable security layer. Invisible to users. Central to your security team. Works with any model, EU-hosted or not.
Chat, API, IDE
Analyze · Mask · Route · Log
Any provider
Names, emails, IBANs, phone numbers are detected and masked before reaching any model.
Injection attempts are caught and blocked before they reach the model.
If a request contains sensitive data, it can be rerouted to an EU-hosted model automatically.
Every call, every Guard decision, every routing choice is logged and auditable.
| Time | User | Event | Action |
|---|---|---|---|
| 14:32:07 | L. Martin | PII detected | 3 fields masked → sent to Mistral Large |
| 14:31:44 | A. Chen | Injection blocked | Request rejected · admin notified |
| 14:31:12 | S. Duval | Rerouted to EU | Sensitive data → rerouted from GPT-4o to Mistral |
| 14:30:58 | J. Park | OK | Passed → Claude Sonnet |
Hosting
Voban connects to any model provider. For sensitive use cases, you can enforce EU-hosted models. For others, use frontier models wherever they run. The Guard brings value either way.
EU-hosted models
Mistral, Qwen, Llama on EU infrastructure. Data stays in Europe. No dependency on non-EU jurisdictions. For your most regulated use cases.
Frontier models
Claude, GPT, Gemini — wherever they're hosted. The Guard still applies: PII masking, logging, injection detection. You get control without giving up capability.
Responsibility
Regulatory requirements are rising. Shadow AI is outpacing your policies. Leadership wants a strategy. Your DPO wants guarantees. Your teams want tools.
Voban is built for those who carry this responsibility: CISOs, DPOs, CTOs, IT Directors. Those who need to open access to AI without losing control.
Product
Users work without friction. Admins define the rules.
Multi-turn conversations, file uploads, admin-configured models. Users work normally. The Guard runs behind the scenes.
OpenAI-compatible endpoint. Works with Cursor, OpenCode, Continue. Same Guard rules apply.
Activate models per workspace. Define Guard rules per team. Set budgets. Track usage by person, team, and model. See every Guard decision.
Early access. Limited spots. First 10 partners in 2026.